Shared resources scheduling using a multi-agent model: the DSCEP framework

Abstract

Recently, multi-agent systems have been successfully applied to the scheduling problem. A new multi- agent framework, called DSCEP (distributed, supervisor, customers, environment, producers), is suggested in this paper. This framework is developed base on the subsistent SCEP models, especial for shared resources scheduling activities. It introduces a dialogue between three kinds of evolved SCEP models leading to a high level of co-operation. It provides a more efficient control of the consequences generated by the local decisions than usual systems for each SCEP model. It also provides different algorithms in order to handle the disturbances occurring at different ranks in manufacturing process. As a consequence, the DSCEP framework can be adapted for various scheduling/planning problems. This model is applied to the shared resources scheduling problem of complex systems, and provide a natural cohabitation between infinite capacity scheduling processes, performed by the multi-site manufacturing orders, and finite capacity scheduling processes, performed by local or remote machines.